Conseal

Latest version: v2025.4

Safety actively analyzes 723954 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

2025.04

Implemented embeddings

- Selection channel
- WS residual-based cost for selection channel

Improvements

- Introduced embedding efficiency to nsF5/F5
- Fixed bug in F5, that prevented shrinkage from occuring.
- Default log-base (2) in simulation avoids second logarithm and is faster now.
- Inverse entropy can take precision now.

Additionally, following was added

- Started `unseal` tool for a basic steganalysis-related forensic analysis of image pairs.

2024.12

Implemented embeddings

- MiPOD
- F5
- S-UNIWARD
- WOW

Improvements

- LSB can now embed with efficiency 2 (by default), but also at the embedding bound

Additionally, following was added

- simulator for binary changes
- distortion-limited sender
- MiPOD LRT for benchmarking of stego security
- coding-related features, e.g., Hamming efficiency and generator of generator of robust Soliton-distributed H matrix

2024.06

Added spatial embedding

- HILL

2024.03

Added first spatial embeddings.

- HUGO
- LSB (replacement and matching, grayscale and color)

2023.12

First release.

Implemented embeddings:

- J-UNIWARD
- UERD
- EBS
- nsF5

API on three abstraction levels:

- high-level (single call to embed)
- mid-level (moderate abstraction, separated cost from simulator)
- low-level (full access to cost computation, probability conversion, and simulator)

24.09

Fixed the source distribution.

© 2025 Safety CLI Cybersecurity Inc. All Rights Reserved.